Best Buy's Dead-Last Mobile Performance: Does It Mean Anything?

Add this to the list of your mobile commerce conundrums. In new analyst rankings, Best Buy tested dead last in all categories, among the major retailers tested. But those results may not matter the slightest in terms of Best Buy's M-Commerce revenue, profits or customer satisfaction.

Why? Because almost no one—including major retailers—is tracking mobile cart abandonment rates. To be explicit, mobile cart abandonment is not exactly the same as its Web counterpart.

In E-Commerce, an abandoned cart generally means someone has spent time on a site, filled up a shopping cart and then stopped the process just before paying. An abandonment in the mobile world, by contrast, is customers who visit a site and then abruptly leave before doing anything significant. At this early stage of M-Commerce, customers who click through to various pages and go so far as to put items in their carts are actually considered successes. After all, many will later consummate those sales on mobile, the Web or in-store.

So mobile cart abandonment raises the distinct possibility that slow, or otherwise flawed, mobile site performance drove customers away. But is that necessarily true? Given the various target audiences for different categories of retailers, is it not realistic to assume that site management has already matched functionality and visuals to target customers?

That's the type of problem behind recent mobile performance rankings being publicized by competing Web performance firms, including Gomez (a division of Compuware) and Keynote.

Take Best Buy, as an example. In the Gomez iPhone application tests covering September 2010, the retailer's performance was dead last in the three categories Gomez tracked: Response Time, Availability and Consistency.

In Response Time, Best Buy's dedicated app delivered a response time of more than eight seconds, compared with the average time of 4.1 seconds and the top-ranked 0.86 seconds from Target. In Availability, Best Buy delivered 97.6 percent uptime, compared with an average of 99.8 percent and a top ranking of 100 percent from six retailers: Macy's, Newegg, QVC, Sears, Target and Wal-Mart.

In Consistency, a low number indicates a well-controlled mobile experience. Best Buy delivered a 10.6 second rating, almost twice as bad as the next weakest performer, which was HSN at 5.7 seconds. The average was 3.5 seconds, and the top score—belonging to Amazon—was 0.7 seconds.

Best Buy fared almost as poorly in the Keynote numbers, appearing second-to-last in overall mobile site availability with a 96.5 percent, beaten for the bottom by the 93.3 percent delivered by In Keynote's mobile availability tests, six retailers achieved 100 percent availability: Wal-Mart, Amazon, eBay, Barcle, and Zeer. (Keynote's list included retailers that are not among the largest.)

The real question, though, is "so what?" Without knowing if those speeds are pushing away customers, the numbers are not especially meaningful.Is Best Buy's site appropriate for what its audience wants? It greets customers with a pleasant holding page that declares "The Best Buy app will be done loading your deals in just a second." It takes more than a second. But for someone who wants to shop at Best Buy, it's not an excruciating wait. The site is heavy with options, but is that necessarily a bad thing? After all, if the customers aren't leaving, there doesn't seem to be a problem.

Matt Poepsel, Gomez's vice president for performance strategies, said that mobile site performance is generally impacted by two very distinct areas. The more visibly obvious area is the size of the download for the homepage and subsequent pages, usually dictated by the number and size of images and colors.

But an equally impactful factor is the number of connections between the mobile device and the Web servers delivering the content. That's generally a matter of third parties, such as advertisers (not a typical M-Commerce issue, but it's there occasionally), Web analytics, rich media services and ratings/reviews companies, Poepsel said.

One problem is that site management often tries to apply lessons learned in E-Commerce to M-Commerce, and that doesn't always work. Consider that issue with the number of hops between the device and the server. Web browsers have gotten much more sophisticated about such matters, while mobile browsers are still in their infancy. Consider, too, the difference between the latest browsers today and the earliest Mosaic browsers from the mid-1990s.

"Today's browsers have become highly evolved. Sites can trick [these modern Web] browsers and they can be made to artificially open up more connections as a performance optimization technique," Poepsel said. "This simply does not work that well in mobile. On the mobile Web, you're really trying to have a minimal number" of hops between the device and the server.

The question of how large a landing page will be well tolerated is also not an easy one to answer, because it varies from category to category. One question that is the same from the Web to mobile: Is it better to have a quick-loading homepage and to place the slower elements deeper within the site (on the rationale that once on the site and clicking, a visitor is committed and less likely to abandon) or to delay the load but have a more powerful initial image and then have a faster experience?

"It comes down to 'Pay me now or pay me later,'" Poepsel said. "It's the difference between Google and Yahoo."